Abstract
A brief meditation on the fate of those currently riding high on so-called artificial intelligence and their opinions on its use and utility.
Engineers that use AI are doing the opposite of engineering#
What happened#
For some reason, when the author checked LinkedIn today to set up an interview for what could turn out to be a pretty cool job he was confronted with this terrible post by some idiot, who is wrong on the Internet. And, being a mature and sensible fellow, the author has become completely immune to idiots being wrong on the Internet, so that’s all for this post.
…
Just kidding! The author is as pedantic and willing to procrastinate more important work than bitching about how stupid other people are than ever!
The post begins with the following line.
Junior engineers should be allowed (or forced, if they don’t want to) to code without AI.
—Some guy that missed the point of the whole AI embroglio to begin with.
Perhaps we could improve on the opening line’s truthiness a bit.
Engineers that use AI are not actually Engineers.
—Some other guy that got the point of the whole AI embroglio to begin with.
It would maybe be better if this didn’t happen#
Where the first version accepts the premise of the marketing term “AI,” which is almost completely unrelated to the actual meaning of the phrase originally coined in the 60’s by the same kinds of nuts that are running around today claiming to be computer scientists and declaring either the end of the world via their general intelligence god or declaring a new golden age via that same general intelligence god, is where it falls flat on its face. That is to say, immediately. The salient fact about marketing (I mean the class of economic actors here) is that they’re very close to the bosses (or owners, they are functionally the same thing), and that means they are lying to you all the time about everything. Despite the author’s empirical nature which would ordinarily require him to present some proof of this claim to you, he’s fairly confident that you already know if you’re even vaguely interested in not being lied to by every source of information in your world.
Which brings us to why the OP is such an idiot. Here is this man, who thinks of himself as at least a “Senior” Developer, claiming that using text extrusion machines is bad for Junior Developers because it prevents them gaining experience and learning new things about how computers work. This means that the OP has managed to get far enough into a career in computing that he’s been given or claimed the title of “Senior” Developer which must have taken at least a few years, yet he has apparently not one single time in that career managed to figure out what his actual job is: to learn new things and apply them to problems that the enterprise needs solved. Ideally, for profit.
The OP is clearly and intentionally set about destroying us all#
Let us, for a moment, set aside the thing that is already well into destroying the “AI Industry”[1], which is the Oroboros Problem[2] and assume that text extrusion is the next big thing that makes programming so simple even an executive could do it. This is, particularly for a “Senior” Developer a pretty serious problem. The owners have never had any love of labor, especially expensive labor, history is littered with examples of bosses preferring to see their labor murdered in defence of their capital, even, many times, champing at the bit for an opportunity to murder their labor seemingly for the sport of it. In the context of the owners being the kinds of monsters that would gladly throw every engineer they currently pay straight onto the street if it mean they could get a pre-pre-yacht to take to their pre-yacht before they finally land at their super yacht for their trip to one of the few bits of functioning ecosystem they’ve cordoned off for their own personal use, this man can only be harmed by the adoption of these ostensible tools. If they live up to the hype, he’s thrown out of his job into a job market that is at least as bad as was the one the author began his career in at the tender young age of 17 in the year 2000 along with all the other “trouble-making” (read good) engineers that were laid off first and it’s a fairly swift spiral down from there to living in his parents basement and joining a men’s-rights group. How can he not see this?
If, as the author (and several noted intellectuals) hypothesize, we’ve seen the best these things could do, they can’t do that anymore, and they’ll never be able to attain even those underwhelming results in terms of productivity loss, then this idiot OP is actively destroying the Internet on which he relies for a living and declaring that we need rules for when people who know less than he does should be allowed to use these amazing and magical machines so that they at least have to learn something before they can give up and sign themselves and the entire Internet over to these things. It’s worth reiterating, if you work in computing, your job is only tangentially related to computers. Your main purpose is to learn in ways that are useful to the enterprise. These things prevent that. Like any other muscle, ones ability to learn atrophies with age and disuse, which leaves one wondering just how exactly, when the extrusion machine fad has passed and the idiot OP finds himself sitting in front of a computer that needs programming, will he even be able to figure out how to find the relevant reference manual? Likely not, the same is likely true for almost every developer who’s started pretending to program computers since 2022. Long term, that’s great for the author if he can say housed and employed until the wave breaks, because these people keep running around the Internet setting fire to everything in site with their “code” and the author happens to be in possession of a fire extinguisher in this metaphor.[3] It’s not so great for the idiot OP or anyone unfortunate enough to train under him since they also will fail to grasp not only what makes them valuable (their capacity to learn in ways useful to the enterprise), but will never develop the intellectual muscles to produce sustained learning effort to being with.
At least we’re not those guys#
This leaves computing generally with at least one, perhaps two or three generations of people who have been told that they are programmers or engineers but will have done no programming or engineering. When this cohort starts being given senior titles, or worse, inevitably ends up ascending into management since most companies are completely dysfunctional and will always promote the wrong guy, they will have absolutely no idea what to do about anything precisely at the moment when the entire Internet is in the process of burning itself down. Perhaps this will be an opportunity for those of us that don’t appreciate the theft of the Internet by corporate interests to build something better, perhaps it just means the end of instantaneous global communications period. About this the author has no hypothesis.
Maybe the sector that’s safest against AI job loss is comedy?
—the editorial board
The author believes the Ororobos Problem is the main cause of this degradation, also that there can be no solution to it since it is only humans that can create new knowledge and we can do that (even when text extrusion machines are helpful in doing so) only so quickly. How many people do you personally know that have been sorely in need of a machine-learning-related technology to assist them decoding protein folding in DNA or mRNA, or any of the other abbreviations the authors beloved wife has explained to him enough times that it’s clear he does not and never will truly understand cell biology? The author knows one, that being his beloved wife, and he’s guessing that you also know very few people in this class.
What this means for the usefulness of text extrusion machines generally is that they aren’t, really. Because they can’t create anything new, but only assemble something that is probably close to what’s been asked for using an existing set of tokens, the quality of those tokens matters and it was at its peak the moment the biggest models consumed the last piece of human output. After that point they can only become less and less helpful until eventually the Oroboros Problem is fully expressed and the snake has consumed not only its own tail, but also its head and from then on can only spit out gibberish. This is the authors hypothesis anyway, we’ll see if its falsified at some point in the next five to ten years.